A good research contribution is ALWAYS built upon a very strong foundation of synthesis of literature review / related work.
It can be difficult for Ph.D. students to differentiate between research and non-research. Here I would like to discuss research in terms of contributions. "Contributions" are the key criteria for determining whether your work is worth publishing or not.
There are four key contributions:
1. "New", "non-obvious" artifact: This is kinda obvious. If you introduce new tools/toolset, software/hardware, algorithms, applications, etc., you will be fine here. Just make sure the artifact you created poses some interesting technical challenges. A "new" but "too-easy" artifact is a no-go. For example, simply using SSVEP to turn on/off is a no-no....it has no technical challenge! The scientific nature is of engineering. You are likely to earn this contribution.
- Common mistake: Make some random games or applications for a very specific use without a clear hypothesis, IV, and DV (i.e., you thought making/building something cool is cool but the problem is you never even think deeply about why and how it will work, and whether it's new or considered to be technically challenging or not....all can be solved by reading papers)
2. "New", non-obvious, significant knowledge: This is mostly based on experimental studies that expose certain "new" phenomenon or knowledge. Anyhow, new but "common-sense" knowledge is a no-go. The scientific nature is empirical and rigorous. If you truly discover something new (and not obvious), you earn this contribution.
- Common mistake: Do some experiments without building upon past work (i.e., you thought you are a genius so you simply come up with your own design and use your common sense to judge that it's ok. As a consequence, your results are either invalid or uninsightful)
3. "New" methodology: If you propose some new research methodology or "new approach to tackle certain known problems", you earn this contribution. It is quite difficult for new Ph.D. students...in my opinion but it's doable.
- Common mistake: Create your own methodology without building upon past work (i.e., you thought you are a genius so you simply come up with a new way, without looking back 20 years what people have done). Most often, new methodologies are mostly adaptations of past "already-good" methodologies.
4. "New" theory: This is probably the hardest of all contributions. Basically, you are proposing some grand new theory that changes the way we see things. Likely you not gonna work on this level as a Ph.D. student. It takes ones' lifetime to realize one new theory.
- Common mistake: Create your own theory without building upon past work (i.e., once again, you thought you are a genius so you just ignore past work). To debunk a theory and create a new theory takes one's lifetime and years of research. So not yet. Calm down and take it slowly.
A good paper usually composes of ONLY ONE contribution. Having too many contributions confuse your readers. Similarly, lacking any of this contribution will easily allow reviewers to stamp a rejection on your paper.
Another good observation is the keyword "new", which means that novelty is very important in BCI research. Thus, it is useful to regularly ask "Why is this problem so hard?", "Why is this problem worth solving", "What is new in your contributions?"
I hope this starting point helps you guys better plan ahead for your research.
Note: There are other contributions such as doing a large-scale survey, writing an opinionated piece, doing a replication study, etc., but they are rather uncommon, and I believe they are not suitable for any Ph.D. students to pursue these types of contributions. Ph.D. students should focus on the basics.